Subclassification of unipolar depression has proven elusive. One explanation may be that use of heterogeneous and overlapping groups has yielded low power studies. Psychopharmacologic studies involving approximately 400 patients suggest that locally defined atypical depression identifies a subgroup of unipolar affective disorder which is distinct from melancholia. Demonstration that atypical depression and melancholia are also familially distinct would add to the validity of these two syndromes. Over three years, this project will evaluate 75 probands with atypical depression and 75 probands with melancholia together with their first degree relatives and spouses. The questions this study will answer are whether these two disorders are familial; if so, whether they are distinct familial disorders, and whether they "breed true". Evaluations of relatives will be performed by raters blind to the probands using the SADS- LA-R (a modification of the SADS which makes DSM III-R diagnoses) for direct interview, and the FISC (a modification of the FH-RDC which makes DSM III-R diagnoses) for uninterviewed relatives. Best Estimate Diagnoses will be made by expert raters based on this information and blind to other family members and probands. Analyses using age-corrected morbid risks and lifetime risk will include MANOVA, survival function, proportional hazards and log linear models. Comparisons will be between families of atypical depressives and relatives of melancholics, as well as between these two groups and six existing control groups -- families of anxious probands and their well acquaintances (from Abby Fyer, M.D.), acquaintances and families of depressed probands (from Jean Endicott, Ph.D.) and community controls and families of depressed probands (from Myrna Weissman, Ph.D.). These latter comparisons will help to establish the generalizability of the results.